RECONNAISSANCE IN CENTRAL TURKESTAN. 
213 
climate. The load of the streams was correspondingly decreased, and they cut 
downward nntil the relief was like that which is represented by the outer valley cut 
in the silt (B, fig. 147). At the same time the front of the old silts was being 
gnawed back and the slope was being prepared on which the terraces (fig. 146) were 
later cut. The surface of the old lake deposit must originally have presented an 
unbroken slope (I H in fig. 146). The bed of the stream, the lower dotted line, 
I B H, must have presented an equally smooth although more concave slope. The 
two must ha\e met at the lake shore when the water was at its lowest level. Even 
the most cursory inspection of the valleys and spurs shows that this point of meet- 
ing must have been far out in the swamp close to the present shoreline. Therefore 
the lake must have been small and shallow, and the climate must have been similar 
to that of to-day or possibly even drier. 
Fig. 148. — The gravel-covered lacustrine deposits at Dungsugot, illustrating the same features as the 
preceding diagrams. For explanation see fig. 146. 
The next changes were those recorded in the terraces and in the graxel which 
fills the valleys represented in fig. 147. It is easy to infer that the lake must have 
risen and fallen twice, and that each rise was associated with the fonnation of a 
terrace and with the partial filling of the valleys with gravel, but of this we have 
as yet no direct evidence. The ne.xt rise of the lake of which we have positive 
proof was probably also the last. It deposited the recent silts (G, figs. 146 and 
148), which rise to a height of 100 feet and bur)- not only the foot of the tenaces, 
but also the isolated hills on which well up the springs of sweet water in the midst 
of the salt swamp. Now a last change of climate has again reduced the lake to verj- 
small proportions. 
